A bill of materials (BOM) is the complete, structured list of raw materials, components, subassemblies, and quantities needed to build one unit of a product. It is the recipe the whole plant plans against, buys against, kits against, and costs against.
Everything downstream of engineering reads the BOM. Purchasing buys to it, production planning explodes it into material requirements, the storeroom builds kits from it, accounting rolls product cost up through it, and quality traces lots through its structure. That is why a BOM error never stays where it started. This post covers the anatomy, the structures, the EBOM/MBOM split, and what a wrong BOM costs, using one example throughout: an industrial utility cart, part number CART-100.
What information goes on a BOM line?
Every BOM line identifies one component and states how much of it goes into one unit of the parent. Seven fields do most of the work:
| Field | What it is | Why it matters |
|---|---|---|
| Part number | The unique identifier for the component. One part, one number. | Free-text or duplicate numbers are how the same bolt gets bought under three names. |
| Description | The human-readable name, written to a consistent convention. | The floor reads this, not the number. “Caster, swivel, 5 in” beats “wheel.” |
| Quantity per | How many go into one unit of the parent. | Every purchase order and kit downstream multiplies from this figure. |
| Unit of measure | Each, feet, pounds, liters. | A line maintained in feet and bought in meters quietly corrupts every count it touches. |
| Reference designator / find number | Where the part goes, keyed to the drawing. | Lets an assembler match the line to a location without asking. |
| Level | Depth in the product structure; level 0 is the finished product. | Tells planning what is built versus bought at each stage. |
| Scrap factor | Expected process loss, stated as a percentage. | Leave it off and you under-buy by the scrap rate on every single order. |
What is the difference between a single-level and a multi-level BOM?
A single-level BOM lists only the immediate children of one parent item. A multi-level BOM, also called an indented BOM, explodes the entire product structure level by level, down to purchased parts and raw material.
For the cart, the single-level BOM at level 0 shows five lines: a frame weldment, a deck assembly, two swivel casters, two rigid casters, and a handle assembly. It does not tell you that the frame weldment consumes steel tube and gusset plates. Each subassembly carries its own single-level BOM, and nesting them produces the multi-level view:
Planning works this structure from the bottom up. Lead times stack down the levels, which is why a missing level-3 lock nut can gate a level-0 ship date.
What is a phantom BOM?
A phantom is a structural level that exists on paper but is never built or stocked as its own item. The hardware kit in the diagram is one: engineering groups the bolts and nuts under HW-123 to keep the deck BOM readable, but the floor never assembles or stores a hardware kit. Planning blows through the phantom and issues the bolts and nuts straight to final assembly.
What is the difference between an EBOM and an MBOM?
The engineering BOM (EBOM) describes the product as designed, grouped by function. The manufacturing BOM (MBOM) describes the product as built: reorganized to match the routing, with the packaging and consumables production actually uses added in.
On the cart, the EBOM groups parts into structure, mobility, and ergonomics, the way the designer thinks. The MBOM regroups the same parts by operation: weld, paint, assemble, pack. It also adds items the EBOM never mentions, because no designer specifies stretch wrap: weld wire, powder coat, the carton, the label.
Plants that run purchasing straight off the EBOM discover the gap the hard way: finished product sitting on the dock with no cartons to put it in. The MBOM is where packaging and consumables get planned like everything else.
What does a wrong BOM actually cost?
A wrong BOM turns into scrap, line-side shortages, and numbers nobody trusts, usually in that order. These are typical failure modes, not statistics; anyone who has lived with a messy BOM will recognize them:
- Wrong-revision builds. Engineering released revision D two weeks ago; the floor is still building to the revision C traveler. Every unit built since the release is rework or scrap, and the material already cut to old dimensions goes with it.
- Shortages discovered at the line. The BOM says two casters; the cart takes four. The kit runs dry mid-shift, the line waits, and the storeroom expedites replacements at freight rates that erase the order's margin.
- Purchasing to a stale BOM. Buyers keep ordering a bracket engineering deleted months ago. Obsolete stock piles up while the part that replaced it runs short.
- Inventory that never reconciles. If completions backflush inventory using BOM quantities, a wrong quantity-per or missing scrap factor drifts the records a little further with every unit built. Cycle counting fights a moving target and loses.
- Corrupted costing. Cost roll-ups multiply the same wrong quantities into standard cost, so margins are misstated and new quotes are priced off fiction.
- Broken traceability. Lot genealogy is recorded through the BOM structure. If the structure is wrong, the question “which material lots went into this unit” has no reliable answer, and that is exactly the question a recall asks. See traceability in manufacturing.
Revision control is the defense, and the discipline is not complicated. Every change goes through an engineering change order (ECO): a written change request; an impact review that checks open purchase orders, on-hand stock, work in process, and cost; an effectivity date stating when the new structure applies; and a disposition for old-revision material: use up, rework, or scrap. The alternative is someone quietly editing a spreadsheet, and every failure mode above follows from that.
The traceability stakes are also turning regulatory in some industries. The FDA’s food traceability rule under FSMA Section 204, finalized in November 2022, requires companies handling foods on the Food Traceability List to keep records of critical tracking events and lot codes, with the compliance date extended to July 20, 2028 (FDA, FSMA final rule). Lot-level traceability of that kind is only as good as the product structure it is recorded against: the BOM is the map a recall follows. Food plants can start with our FSMA 204 overview.
How do you clean up and control a BOM?
The sequence below works because it fixes the data first, then the process that protects the data:
- Audit the BOM against what the floor actually builds. Take the multi-level BOM to the line and watch a unit go together. Every operator workaround, penciled correction on a traveler, and “we always grab two extra” is a BOM error announcing itself.
- Fix quantities, units of measure, and scrap factors in one pass. Correct quantity-per from observed builds, standardize units of measure, and set scrap factors from real usage history instead of guesses.
- Give the BOM one owner and route every change through an ECO. One named person approves structure changes. No direct edits; every change carries an effectivity date and a disposition for old-revision material.
- Align the EBOM to an MBOM. Restructure to match the routing and add packaging and consumables, so purchasing sees everything production consumes.
- Sync the corrected BOM to kitting and costing. Rebuild pick lists and re-run cost roll-ups so the ERP the storeroom, and finance all read the same structure.
- Put a recurring accuracy audit on the calendar. Sample a few builds each month and score the BOM line by line, the way you would cycle-count inventory. Accuracy decays without maintenance; the audit is the maintenance.
Most BOM errors surface first as scribbles: a corrected quantity on a paper traveler, an operator note that the drawing is wrong, a kit shortage logged nowhere. Harmony is an AI-native layer that connects the ERP, spreadsheets, and that paperwork into one operational layer, with no rip-and-replace. Paper logs and checklists become data captured on tablets, so a deviation noted at the line is searchable instead of lost, and inventory intelligence flags shortages before they hit the line. CLS replaced paper-based production logging with real-time operational intelligence on the same principle: the floor already knows where the BOM is wrong, and the fix starts with capturing what it knows.